Trump Wants a Piece of Big AI: Inside the White House Plan to Trade Policy for Equity in America's Most Valuable Companies

President Donald Trump announced on June 5, 2026, that he is in active discussions with the leaders of the largest artificial intelligence companies about having the U.S. government acquire "pieces" of their firms [1]. The talks — which follow the administration's 2025 deal to take a 10% stake in Intel — would bring top executives from OpenAI, Google DeepMind, Anthropic, xAI, and others to the White House to negotiate the terms under which hundreds of billions of dollars in domestic AI investment might flow, and what Washington would get in return [2].

The proposal centers on having AI firms voluntarily cede shares to the federal government. The resulting equity could be held in a sovereign-wealth-style vehicle, with dividends eventually distributed to American households [3]. OpenAI CEO Sam Altman first pitched the concept directly to Trump in early 2025 and was in Washington meeting with lawmakers and administration officials in early June 2026 [3]. Anthropic, by contrast, has said it is not in discussions about providing equity to the government [2].

The Scale of Capital at Stake

The sums involved dwarf anything seen in prior White House business summits. Aggregate annual AI infrastructure spending by the five largest U.S. cloud and technology companies rose from roughly $230 billion in 2024 to $380 billion in 2025 and is projected to reach $660–690 billion in 2026 [4][5].

Source: Company earnings reports, Fortune, Futurum Group

Data as of Jun 1, 2026CSV

Amazon leads with a projected $200 billion in 2026 capital expenditures, up from $131 billion in 2025. Google follows at $175–185 billion, up from $91 billion [4]. Beyond the hyperscalers, the Stargate joint venture — involving OpenAI, Oracle, and SoftBank — has pledged up to $500 billion for AI data centers and power generation [6]. Anthropic's revenue run rate surpassed $9 billion by January 2026, up from roughly $1 billion at the end of 2024, while OpenAI ended 2025 with approximately $20 billion in annual recurring revenue [4].

Overseas deployment figures are harder to pin down, but several companies have announced major international projects. Microsoft committed $80 billion to AI-enabled data centers globally in fiscal 2025, and Google has announced builds in Malaysia, Thailand, and the United Kingdom [4]. The White House framing positions the equity-stake discussions as a mechanism to tilt the balance further toward domestic spending.

What the Industry Wants in Return

The June 2, 2026, executive order — titled "Promoting Advanced Artificial Intelligence Innovation and Security" — offers a window into the policy concessions industry has already extracted [7]. After months of internal administration debate and direct lobbying from tech executives, Trump signed a version that establishes only a voluntary framework for pre-release government review of frontier AI models [8]. Companies would provide the government early access to models for up to 30 days before release, but the order explicitly states that the process creates no licensing, permitting, or approval requirements [7][9].

This represents a significant retreat from the national security establishment's preferred approach. Anthropic's April 2026 announcement that it was restricting release of its Mythos Preview model — because of the model's ability to identify and exploit software vulnerabilities — had prompted alarm across both Silicon Valley and Washington [8]. The voluntary framework that emerged reflects industry's position that mandatory safety reviews would slow innovation and push development offshore.

On export controls, the administration has moved toward relaxing restrictions on AI chip sales to allied nations, a long-standing industry request [9]. Visa expansions for skilled AI workers and antitrust forbearance — particularly relevant as OpenAI, Anthropic, and SpaceX prepare IPOs — remain active topics in the discussions, though no formal policy changes have been announced [2][3].

The Intel Precedent

The template for the current AI negotiations was set in August 2025, when the government acquired a 9.9% stake in Intel — 433.3 million shares at $20.47 per share — using $8.9 billion from unspent CHIPS Act allocations and Department of Defense Secure Enclave funds [10][11]. The government holds passive ownership with no board representation but can vote on matters requiring shareholder approval [10].

Intel's stock has roughly doubled since the deal, giving the government's stake a paper gain of approximately $47 billion [12]. This outcome has emboldened administration officials to pursue similar arrangements with AI-native companies. But the Intel deal also drew scrutiny: Senator Elizabeth Warren pressed Commerce Secretary on the terms, questioning whether the arrangement amounted to the government subsidizing a company's stock price while taxpayers bore the downside risk [13].

Do These Pledges Actually Materialize?

The historical record of White House investment announcements is mixed. Trump has claimed over $20 trillion in new U.S. investment commitments during his current term, but CBS News reported that the numbers "don't add up" [14]. A Bloomberg Economics analysis found that of $9.6 trillion in listed pledges, only approximately $7 trillion constituted genuine investment commitments — the remainder were trade purchase agreements or other transactions that do not represent capital investment [14].

The CHIPS Act offers a more granular case study. Biden credited the legislation with attracting $640 billion in private investment, but PolitiFact found actual commitments were less than half that figure [15]. As of December 2024, the Commerce Department had allocated about $32 billion of the $39 billion in available funds, catalyzing roughly $158 billion in realized construction investment — substantial, but well below headline announcements [16]. More troubling, GlobalFoundries announced $500 million in share buybacks in 2026 after receiving up to $1.5 billion in CHIPS Act funding, while Intel laid off more than 15,000 employees in 2024 even as its federally backed fab construction continued [16].

The pattern suggests that announced figures reliably overshoot actual deployment, and that even when investment materializes, the jobs and economic benefits may not follow the script presented at signing ceremonies.

Jobs: Real Numbers Behind the Promises

Proponents of the AI buildout point to significant employment effects. Permanent data center employment is projected to reach 650,000 positions by 2026, a 30% increase from 2023 [17]. Construction of nearly 2,800 new data centers is expected to generate an estimated 4.7 million temporary construction jobs [17].

Source: Brookings, IndexBox, AFCOM

Data as of Apr 1, 2026CSV

Many of these roles do not require a four-year degree. Demand for robotic technicians grew 107% between 2022 and 2026, while HVAC/cooling engineer openings rose 67% and industrial automation technician postings increased 51% [18]. Data center engineers earn $84,000 to $196,000, with senior roles reaching $240,000, and power electronics specialists command $150,000 to $250,000 as companies compete for scarce talent [18]. Nvidia CEO Jensen Huang has argued that the buildout's biggest beneficiaries will be blue-collar tradespeople, not white-collar knowledge workers [19].

But the industry simultaneously faces a skilled worker shortage. Analysts estimate the global sector will need more than 200,000 additional electricians, technicians, and project managers by 2026 [20]. Whether the labor market can supply workers at the pace demanded by a $690 billion annual spending trajectory remains an open question.

Ethics, Conflicts, and the Sovereign Wealth Question

The structure of these negotiations has drawn criticism from governance watchdogs. A sitting president is personally convening private talks over federal policy with companies whose valuations could be directly affected by the regulatory and investment decisions under discussion [2]. The Stargate venture, announced at the White House in January 2025 with Trump standing alongside Altman, SoftBank's Masayoshi Son, and Oracle's Larry Ellison, benefited from an executive action using emergency declarations to fast-track permitting for participating companies [6].

Senator Bernie Sanders has proposed the American AI Sovereign Wealth Fund Act, which would impose a one-time 50% tax, paid in shares, on OpenAI, Anthropic, and xAI, depositing the equity into a public fund with citizen voting rights and board representation [3]. This approach differs from the administration's voluntary framework in that it would compel rather than invite participation, and would grant the public governance rights rather than passive ownership.

The legal pathway for the voluntary approach remains unsettled. How a share transfer from private companies to the federal government would be structured under existing securities law, and what authority the executive branch has to negotiate such arrangements without congressional authorization, are questions that have not been publicly addressed [3].

The Case for Government Coordination

Critics of the meeting are not necessarily right that it represents only cronyism or regulatory capture. There are models of successful government-industry coordination in technology sectors.

South Korea committed 8.6 trillion won ($6.5 billion) in 2026 alone to support AI, semiconductors, and quantum computing through a multi-agency strategy, and unveiled a broader 700 trillion won ($534 billion) plan to shift from memory-dominated chipmaking to a more diversified semiconductor ecosystem [21][22]. The plan emphasizes cross-ministry coordination, grouping 513 technologies into 19 common categories to streamline regulatory and investment alignment.

France's National Strategy for AI, embedded in the broader France 2030 investment plan, deploys a national AI coordinator who reports to the prime minister and convenes a steering committee of eight ministries monthly [23]. The approach has helped France attract significant AI infrastructure investment, including partnerships with Nvidia for sovereign computing capacity.

The argument for centralized coordination is straightforward: market-driven allocation alone may underinvest in shared infrastructure (grid capacity, workforce training, research) and overconcentrate benefits in a small number of firms and geographies. A government that holds equity in the companies it regulates has a direct financial interest in their success — which is either an alignment of incentives or a conflict of interest, depending on one's perspective.

Who Pays the Price

The communities hosting AI infrastructure bear costs that rarely feature in White House announcements. Large data centers consume up to 5 million gallons of water per day — equivalent to the daily use of a town of 10,000 to 50,000 people [24]. In Northern Virginia, the roughly 200 operational data centers in Loudoun County used around 900 million gallons of water in 2023, a 63% increase from 2019 [24].

Utility ratepayers nationwide saw their electric bills increase at roughly twice the rate of inflation in 2025, with over $60 billion in rate increases approved countrywide [25]. Utilities are increasingly using Construction Work In Progress (CWIP) — a financing mechanism that charges customers for construction costs during the building phase — with at least 40 states now permitting it, roughly double the number from a decade ago [25].

Community opposition has had real effects: between March and June 2025, local resistance led to $98 billion in data center projects being blocked or delayed [25]. In Memphis, Tennessee, residents and the NAACP filed a notice of intent to sue under the Clean Air Act over xAI's Colossus data center, which is installing more than 30 natural gas turbines [25]. In 2026 alone, lawmakers in more than 30 states have introduced over 300 bills addressing data center siting, energy use, and cost allocation [25].

The Energy Equation

The electricity demand projections are stark. U.S. data center power consumption is expected to nearly double between 2025 and 2028, from 80 to 150 gigawatts [26]. By 2030, the country is on track to consume more electricity for data centers than for the combined production of aluminum, steel, cement, chemicals, and all other energy-intensive manufacturing [27].

Source: IEA, Goldman Sachs, Pew Research

Data as of May 1, 2026CSV

Globally, data center electricity consumption is projected to reach 945 TWh by 2030, roughly 2.5 times the 2024 level of 415 TWh [27]. The International Energy Agency projects that electricity demand from AI-optimized data centers specifically will more than quadruple by 2030 [27].

What powers this growth matters for U.S. climate commitments. Goldman Sachs analysis projects a 160% increase in data center power demand by 2030 [27]. While some companies have committed to sourcing renewable energy, the sheer scale of new demand means natural gas generation will fill much of the gap, at least in the near term. The Memphis xAI facility's reliance on gas turbines illustrates the tension: the fastest path to powering AI infrastructure often runs through fossil fuels [25].

Twenty-seven states have enacted or are advancing legislation requiring data center developers to cover energy costs and report usage, with California, Ohio, and Utah going beyond the federal government's voluntary Ratepayer Protection Pledge [25]. But no federal framework currently ties AI investment incentives to carbon emission performance.

What Comes Next

The White House meeting, when it occurs, will test whether a government can negotiate itself a share of the AI economy's upside without compromising its role as a neutral regulator. The Intel deal provides an encouraging financial return but an uncertain governance template. The voluntary nature of the proposed arrangements makes them dependent on the continued goodwill of companies whose market power gives them significant bargaining leverage.

The $690 billion in projected 2026 AI capital spending is real money being deployed at an extraordinary pace [5]. Whether it produces broadly shared prosperity or concentrates benefits among a handful of firms while distributing costs — higher utility bills, strained water supplies, displaced communities — across millions of households will depend on details that remain far from settled.

Sanders' sovereign wealth fund proposal and the administration's voluntary equity approach represent two poles of a debate that is only beginning: who should own the economic surplus generated by artificial intelligence, and on what terms [3]. The answer will shape not just U.S. technology policy but the social contract around the most capital-intensive industrial buildout since the postwar era.